AI for Writing a Soundwalk Listening Score

As of 2026-10-16 · AZET 운영팀 (the AZET team). We operate azet.io and publish this guide for fun and reference.

"Listen for the bell at the corner" is a fragile instruction when there may be no bell. We prefer a listening score that leaves room for the place to sound different from the plan. Here, a score means a short sequence of instructions for directing attention, not musical notation.

The route, observations, and score excerpts in this guide are fictional. If you use an AI text assistant, provide your own descriptions of places and sounds. The assistant is being asked to help phrase an exercise, not to hear the route, inspect access, or report what happened there.

Give the score an arc

For a short soundwalk, choose a beginning, a change of attention, and a closing note. These can be three stopping points you already know and personally choose. The score should not need a particular sound to occur before the listener can continue.

Our fictional maker supplies this brief:

Three stationary listening pauses along a familiar courtyard route: doorway, middle bench, covered passage. These are placeholders for a writing exercise. At the doorway I want attention on starts and stops. At the bench I want attention on overlapping sounds. In the passage I want a comparison with the previous pause. No recordings or source identification required.

That brief describes a desired experience. It does not describe sounds already heard. Keeping it in the future tense helps prevent the draft from becoming a fabricated field report.

Ask for actions rather than atmosphere

A prompt could read:

Draft a three-part listening score from my supplied brief. Give one main attention task at each pause and a short note-taking instruction afterward. Make every task usable if an expected sound is absent. Do not invent sounds, identify sources, or describe results. Keep movement separate from stationary listening. Label the entire sample hypothetical.

Here is a hypothetical score written for that brief:

Doorway: beginnings. At your chosen stopping place, attend to the start and end of one sound you notice. Write a few words about its pattern. If no distinct beginning or ending stands out, write that instead.

Bench: layers. At the next stopping place, notice whether you can distinguish overlapping sounds. Describe their relationship in your own terms. If one sound dominates your attention, describe that experience without forcing a second layer.

Passage: return. At the final stopping place, choose one quality you wrote about earlier and consider whether you notice it here. Record a similarity, a difference, or an inability to compare.

Close. Circle one phrase you would keep and underline one phrase you would revise after reading your notes.

This score has an order without requiring a predetermined soundtrack. The listener supplies the content. The instructions supply a way to attend to it.

Repair a cue that scripts the answer

Imagine a hypothetical draft adds: "Hear the distant footsteps growing louder under the roof. Describe their echo."

The human editor should remove the assumed footsteps, distance, change in loudness, and echo. None appears in the brief. A replacement could be: "Choose a sound you notice at this stopping place. Describe whether it seems steady or changing to you."

That wording still offers a task. It simply does not prewrite the result. If a listener later records "footsteps, perhaps," preserve the qualification in their notes instead of turning it into a confirmed source.

Another weak cue is "listen deeply." It may set a tone, but it does not tell the participant what to do next. Add a concrete action, such as choosing one sound and writing two descriptive words, while leaving the words themselves to the listener.

Rehearse the instructions on paper

Before using a score, read each line as though the expected sound never arrives. Check whether the listener still has a clear next step. An absence can be a legitimate entry; it should not strand the exercise.

Then check the number of simultaneous tasks. "Notice rhythm, distance, pitch, texture, source, and emotion" asks for six decisions at once. For this fictional score, keep the selected focus at each stop and put optional comparisons at the end.

Leave space between the cues for the listener's own writing. Afterward, review the instructions against the actual notes they provide. Revise any cue that was confusing, but keep that revision separate from what was heard during the walk.

FAQ

Does a listening score need timings?

Only if timings serve your intended exercise. You can let a participant decide when to move on, or supply a duration as a proposed constraint. Label it as part of the plan, not as a duration anyone has already completed.

What if the place is quieter than expected?

Use the score's alternate instruction: record that no distinct event or layer stood out. Do not add imagined sounds to make the page fuller. A later revision might simplify the cue so it does not depend on a busy soundscape.

Should listeners name every sound source?

Not for this exercise. Descriptions such as "short repeated clicks" can be enough if those are the listener's own words. If they guess a source, mark the guess separately. The score can stay focused on listening patterns rather than identification.

Our dated product reference

The maker introduction dated 2026-10-03 describes azet as an AI assistant and an email early-access waitlist. This guide does not document an AZET audio or route feature.

For the waitlist described in that dated introduction, visit https://azet.io and check the website for the latest availability.